Abstract:

OBJECTIVE: To explore the effect of different convergent conditions on accuracy of simulation results from a three dimensional finite element model of the pelvic ring.
METHODS: A first-order linear load of 600 was applied on the S1 vertebral endplate in an established three-dimensional finite element model. The step length was set to 0.1 s. The boundary condition was set as constraint of 6 degrees of freedom in the proximal femur. Static and dynamic explicit convergences with 6 different weight scale factors were calculated retrospectively, and all the simulated results were compared with the experimental results in order to verify the accuracy.
RESULTS: The static convergence predicted most accurate with the linear regression coefficient 0.88. With the increase of weight scale factor, the time cost decreased. However, the accuracy of the predicted results decreased. There was statistically difference between the simulation results and experimental results when the weight scale factor achieved 3 000 (P < 0.05) and the coefficient of linear regression was lower than 0.8.
CONCLUSION: It suggested that as for the complex finite element model, especially when the model contains complex contact conditions, dynamic explicit convergence can be an alternative solution to static convergence if the latter failed. Also proper weight scale factor should be used to decrease the time cost under the condition that the error was in the limited.